Exercise Capacity and Perceived Exertion on Treadmill Stress Test With vs Without Surgical Mask: a Randomized Clinical Trial in Healthy Adults.

Background: The graded exercise stress treadmill test (GXT) is among the most frequently performed tests in cardiology. The COVID-19 pandemic led many healthcare facilities to require patients to wear a mask. This study evaluated the effect of the surgical face mask on exercise capacity and perceived exertion.
Methods: In this prospective, randomized crossover trial, 35 healthy adults performed a GXT using the Bruce protocol with a surgical mask, and without a mask. The primary outcome was exercise capacity in metabolic equivalents (METs), and the secondary outcome was exercise perception on the modified Borg scale (from 0 to 10). Effort duration, heart rate, oxygen saturation, and blood pressure were also analyzed.
Results: Exercise capacity was reduced by 0.4 MET (95% confidence interval [CI] -0.7 - -0.2) during the GXT with a mask (11.8 ± 2.7 METs vs 12.3 ± 2.5 METs, p = 0.001), and the final perceived effort increased by 0.5 point (CI 0.2 - 0.8; 8.4 ± 1.3 vs 7.9 ± 1.6, p = 0.004). Effort duration was cut down by 24 seconds (CI -0:39 - -0:09; 10:03 ± 2:30 vs 10:27 ± 2:16, p = 0.003). Oxygen saturation was slightly lower at the end of the test with the mask. There was no significant difference in heart rate, or blood pressure during the test.
Conclusion: Wearing a surgical mask causes a statistically significant decrease in exercise capacity and increase in perceived exertion. This small effect is not clinically significant for the interpretation of test results.

INTRODUCTION

Exercise testing is a convenient and valuable tool frequently used to assess patients with known or suspected cardiovascular disease. The most acknowledged indication for graded exercise stress test remains the diagnosis of coronary artery disease. However, exercise testing is also frequently used for prognostic evaluation of different cardiopathies and risk stratification in cardiac rehabilitation.

Since March 2020 and the declaration of a global pandemic caused by COVID-19, many healthcare facilities request that patients always wear a mask on the premises, to prevent the potential spreading of the virus to other patients or workers. In 2020, the American Society of Echocardiography published a statement on the reintroduction of echocardiographic services during the pandemic recommending that, due to the potential aerosol production during stress testing, patients and providers should consider wearing minimally a surgical face mask, and a N95 in particular circumstances. In our university-affiliated hospital, patients who need a graded exercise stress treadmill test (GXT) must wear a surgical face mask during the test. Many of them reported that the test was more difficult with a mask and believed that they would achieve a better result without it. As exercise capacity is an important predictor of mortality , and a useful parameter in the follow-up of many patients with cardiovascular diseases, it is essential to determine if wearing a mask affects the results of GXT and if adjustments to the test interpretation are warranted.

Before the COVID-19 pandemic, a few studies evaluated the effect of wearing a surgical mask on exercise testing results. One study included 20 healthy adults undergoing a 1-hour treadmill walk (5.6 km/h) and found no clinically significant physiological change. However, another study of 44 healthy participants found that wearing a surgical face mask increased dyspnea during a 6-minute walk test, without changing the distance performed or other physiological parameters. Many more studies on this subject were performed since 2020, investigating cloth, surgical, and N95 face masks during vigorous ergocycle effort. The results were variable but globally suggested a mild effect or no effect at all on exercise capacity despite a small deterioration of the pulmonary function parameters7, 8, 9.

When the current study was designed, there was no published data assessing the effect of wearing a face mask during a GXT using the Bruce protocol. Because it is the most commonly used protocol for treadmill testing in cardiology, it appeared important to assess the potential effects of the mask and their impact on the interpretation of results. More recently, Driver et al. demonstrated a 14% reduction in exercise duration and a 29% decrease in maximal oxygen consumption during a GXT using the Bruce protocol with a cloth face mask.

The purpose of the present study was to evaluate if wearing a surgical face mask affects exercise capacity and perceived exertion during a GXT using the Bruce protocol in healthy participants. The mask type and the protocol were chosen to reflect the most common “real-life” setting in our cardiology laboratory. Based on the literature about surgical face masks and exercise testing, we hypothesized that the effect would not be clinically significant.

MATERIAL AND METHODS

Population

Thirty-five healthy adults between 18 and 65 years of age who had received at least one dose of COVID-19 vaccine, at least 4 weeks before their first visit, were recruited. Exclusion criteria included: proven or suspected cardiac or respiratory disease or other medical condition, requiring medication for the treatment of cardiovascular, respiratory, or neuromuscular conditions, walking difficulties limiting the capacity to perform a GXT, contra-indications to GXT, abnormal resting ECG limiting the interpretation of the stress test, incapacity to wear a face mask or to understand and follow instructions. To reduce the risk of COVID-19 spread, participants who had symptoms related to COVID-19, an unprotected exposition to a COVID-19 positive person, or who had traveled outside the country in the 14 days before their research appointment had to cancel or postpone their participation.

The recruitment was done by sending emails to our organization's employees and through posters placed in hospitals.

Study Design

The study protocol was approved by the university research ethics board of the Sherbrooke University Hospital Center (CHUS). It was registered on ClinicalTrials.gov (NCT04891120). The study design was a randomized counter-balanced crossover trial. The participants were randomized into 2 groups. The participants in group A performed their first GXT with a surgical face mask and those in group B started the study without a mask. On the second visit, all the participants crossed over to the other intervention. The second visit was scheduled at least 48 hours after the first one. If possible, both GTX were scheduled at the same time of the day. Participants were instructed to reproduce the same hydration, nutrition, and resting conditions during the 2 visits. They were asked to refrain from strenuous activity for 48 hours before each test.

Outcomes

The primary outcome was change in exercise capacity (measured in METs) for participants undergoing the GXT with vs without a surgical face mask. The secondary outcome was the difference in perceived exertion on the modified Borg scale. We also compared different exercise parameters with and without a surgical mask: heart rate, blood pressure, oxygen saturation, the time before reaching 85% of maximal predicted heart rate, and the main reason why the participant had to stop the test.

Interventions

Informed consent was signed by all participants before beginning the study in accordance with the Declaration of Helsinki. On the first visit, the eligibility criteria were confirmed. The characteristics of each participant were collected: age, sex, weight, height, body mass index, and self-reported mean duration of physical activity per week. A physical examination was performed to rule out any contra-indication for a GXT, followed by a resting ECG.

The GXT was supervised by an electrophysiology technician and a 5th-year cardiology resident. Standardized instructions were given to the participants and no physical or verbal encouragement was allowed. The modified Borg scale was explained to the participants (0 was described as sitting and doing nothing and 10 as the most difficult exercise they could do), and they could refer to a visual chart placed in front of the treadmill. Participants were not allowed to use the handrails of the treadmill to help with their effort. The standard Bruce protocol was used for all participants. This protocol is composed of 7 stages with increasing inclination and speed between each step. The stages last 3 minutes, except for the last one which can continue with the same parameters until the participant reaches exhaustion. Blood pressure was obtained at rest and during exercise with a manual sphygmomanometer (HillromTM, Welch Allyn FlexiPort Reusable Blood Pressure Cuff, adult size), heart rate was measured with the treadmill ECG, and saturation was obtained with a portable signal extraction pulse oximeter (Masimo Rad-5) before starting the test and during the effort. We used a T2100 treadmill from General Electric Healthcare. After 1 minute of each stage and at the end of the GXT, blood pressure, heart rate, oxygen saturation, as well as perceived exertion (modified Borg scale from 0 to 10) were collected. The participants were instructed that the speed and incline would increase 10 seconds before the beginning of each stage. They were instructed to perform the maximal effort possible and let the technician know when they could no longer continue for the test to be stopped. The GXT could also be stopped if, for any reason, the supervisors deemed that pursuing the test was dangerous. The reason that led to the termination of the GXT and the time (min: sec) at which participants reached 85% of their maximal predicted heart rate (85% of 220 – age; an important target during stress testing in clinical cardiology) were collected, as well as the effort duration (min: sec), the maximal heart rate, and the METs achieved. The METs were estimated with the American College of Sports Medicine equation: VO2 = 3.5 + (0.2 × speed) + (0.9 × speed × % grade). A monitored sitting rest period of 3 minutes was completed. The treadmill’s screen displaying information about the test was hidden from the participants, to prevent them from getting access to any information that could affect the results of the tests.

The masks used were blue surgical masks with ear loops, splash resistance of >16 kPa, and a bacterial filtration power of 98% (PRI-MED Medical Products, Alberta, Canada). A strict disinfection procedure was applied after each GXT and the supervisors wore a gown, a protective visor, and gloves during each test.

The resting and stress ECGs were analyzed by a cardiologist. If an anomaly was observed, the participant was referred to the appropriate resource for further investigation.

Sample Size

For the sample size calculation, a difference of 1 MET was chosen as clinically significant according to the results of prior studies demonstrating a difference in mortality risk with a change of 1 MET in exercise capacity , , . The standard deviation from the reference population used was 2.1, following the normal values for healthy adults available in the American Heart Association Guidelines. The correlation between results without vs with the mask was presumed to be more than 75%. The sample size calculated for an alpha value of 0.05 and a power of 80% to detect a 1 MET difference was 20 based on a paired Student’s t-test. We recruited 35 participants to account for possible losses to follow-up or incomplete data related to COVID-19 restrictions.

Randomization and blinding

A randomized sequence in blocks of 4 was generated. Participants were assigned to group A or B at their first visit following this sequence. Participants, supervisors, and authors were not blinded to the intervention due to the nature of the study, but the statistician was blinded during data analysis.

Statistical method

Demographic quantitative data are presented as mean with standard deviation if normally distributed or as median with interquartile range if non-normally distributed. Demographic categorical variables are presented in frequency and percentage. For the primary and secondary outcomes, a paired Student’s t-test was chosen for normally distributed variables and a Wilcoxon’s signed-rank test was used for non-normally distributed variables. Normality was determined based on visual assessment of the histograms. McNemar’s test was used for dichotomic variables. Equivalence testing using the paired “two one-sided t-tests” (TOST) was also performed to assess whether the exercise capacity (measured in METs) was equivalent for participants undergoing the GXT with vs without a surgical face mask (margin of equivalence of ±1 MET , ). TOST tests the null hypothesis of statistical difference, meaning that if the null hypothesis is rejected, we can conclude that the observed difference is included in the equivalence limits (±1 MET). The results with a p-value < 0.05 were considered statistically significant. IBM SPSS Statistics 27 statistical software and R version 4.1.3 were used.

RESULTS

A total of 35 participants were randomized in the study, of which 33 completed the two GXT and were included in the analysis (Figure 1 ). The recruitment period was May 2021, and the GXT took place in May and June 2021. The trial was stopped when the pre-determined maximal number of participants was reached.

Figure 1

Flow chart of the study. Groups A and B were randomized.

The characteristics of the participants are shown in Table 1 . Participants were mostly females (66%) and aged between 21 and 65 years old.

Table 1

Participants’ characteristics

	N = 35
Sex (male)	12	34%
Age (years)	41.8	±13.3
Height (m)	1.69	±0.09
Weight (kg)	66	(59 – 83)
BMI (kg/m2)	23	(21 – 29)
Weekly self-reported exercise (h)	4	(3 – 8)

Data presented are mean and standard deviation (if normally distributed), median and interquartile range (if not normally distributed), or frequency and percentage (categorical variables).

BMI: body mass index.

Table 2 shows the exercise parameters with and without mask. Figure 2 shows the difference in exercise capacity and perceived exertion. During the GXT without a mask, the mean METs achieved were 12.3 ± 2.5, and 11.8 ± 2.7 with a mask, for a mean difference of -0.4 MET (95% confidence interval [CI] -0.7 – -0.2, p = 0.001, TOST p-value < 0.001). The exercise duration was 10:03 ± 2:30 vs 10:27 ± 2:16 (-0:24, 95% CI -0:39 – -0:09, p = 0.003). At the end of the GXT, the perceived exertion on the Borg scale was 8.4 ± 1.3 vs 7.9 ± 1.6 (+0.5, 95% CI 0.2 – 0.8, p = 0.004). The reason to end the test was more frequently dyspnea with than without a mask (64% vs 36%). Oxygen saturation was slightly lower at the end of the test but remained within expected values (96.2 ± 2.0% with mask vs 97.3 ± 1.6% without mask, p = 0.002) and is not considered clinically relevant. There was no significant difference in heart rate, blood pressure, or the duration before reaching 85% of the predicted maximal heart rate.

Table 2

Comparison of exercise parameters with vs without wearing a surgical mask

	With mask (n = 33)	Without mask (n = 33)	Mean difference	95% confidence interval	p-value
Functional capacity (METs)	11.8 ± 2.7	12.3 ± 2.5	-0.4	(-0.7 – -0.2)	0.001
Effort duration (min:sec)	10:03 ± 2:30	10:27 ± 2:16	-0:24	(-0:39 – -0:09)	0.003
Perceived exertion at the end of the test	8.4 ± 1.3	7.9 ± 1.6	0.5	(0.2 – 0.8)	0.004
Resting HR (bpm)	72.6 ± 12.0	73.1 ± 12.6	-0.6	(-3.8 – 2.7)	0.719
Maximal HR (bpm)	175.0 ± 14.4	177.3 ± 15.9	-2.3	(-5.3 – 0.8)	0.136
Time to reach 85% of maximal predicted HR (min:sec)	7:25 ± 1:59	7:21 ± 2:04	0:04	(-0:17 – 0 :25)	0.683
Resting oxygen saturation (%)	98.7 ± 1.1	98.6 ± 1.3	0.2	(-0.5 – 0.8)	0.623
Oxygen saturation at the end of the test (%)	96.2 ± 2.0	97.3 ± 1.6	-1.2	(-1.8 – -0.5)	0.002
Resting BP (mmHg)	110/76 (105/68 – 121/80)	120/77 (110/70 – 129/80)	-4/-3	-	0.079/ 0.029
BP at the end of the test (mmHg)	161/70 (150/60 – 180/80)	160/72 (150/69 – 177/80)	2/-2	-	0.612/ 0.173
Dyspnea as reason for stopping the test n (%)	21 (64)	12 (36)	-9 (27)	-	0.022

Data presented are mean and standard deviation (if normally distributed) or median and interquartile range (if not normally distributed) or frequency and percentage (categorical variables). Perceived exertion was measured on the modified Borg scale (from 0 to 10).

METs = metabolic equivalents, HR = heart rate, bpm = beats per minute, BP = blood pressure.

Figure 2

Changes in exercise capacity (METs) and perceived exertion (modified Borg scale; from 0 to 10) for participants at the end of the graded treadmill stress test with a surgical face mask vs without a mask. For most participants, exercise capacity decreased, and perceived exertion remained the same or increased. Data is shown as minimum and maximum with median and interquartile range.

DISCUSSION

This study is relevant to the current reality of exercise testing in cardiology during the COVID-19 pandemic. It helps to answer important questions about the effect of surgical face masks on exercise capacity and perceived exertion. It also informs health professionals about possible changes affecting the interpretation of test results.

The design of the study is consistent with real-life exercise testing in our cardiology laboratory, and many other laboratories throughout the world, using the well-known Bruce protocol and a common type of face mask. Consequently, the results apply to GTX with similar surgical face masks but should not be extrapolated to other mask types such as cloth or N95.

The difference of -0.4 MET reaches statistical significance but is below the threshold of 1 MET considered significant for prognostic value , (TOST p-value < 0.001). Similarly, a difference in perceived exertion of less than 1 point on the Borg scale is not deemed clinically meaningful according to studies in patients with chronic pulmonary conditions , .

Our results are partly discordant with those obtained by Driver et al. using the same graded treadmill protocol, and a similar crossover design. They observed a 14% reduction in exercise duration when 31 healthy adults performed a GXT using the Bruce protocol with a cloth face mask compared to without a mask, whereas we observed a difference in exercise duration of less than 4% with a surgical face mask vs without.

Many factors could explain this difference, notably the type and fit of the masks. Driver et al. studied cloth masks that are generally thicker and fit more closely to the mouth than surgical masks. Moreover, they used a metabolic testing (or spirometry) mask on top of the face mask, which blocks air circulation on the sides of the cloth mask. Similarly, Fikenzer et al. tested the effect of surgical and N95 face masks during an ergocycle progressive stress test in 12 healthy men, also using a metabolic testing mask on top of the face mask. They observed a reduction of the VO2 max of 13% with the more fitted N95 mask, and a reduction of only 4% with the surgical mask. This latter difference did not reach statistical significance (p = 0.063). These results suggest that thicker and more fitted masks have a greater impact on exercise testing results. However, other studies such as those of Epstein et al. and Shaw et al. found no significant impact of either surgical, cloth, or N95 masks during cycle ergometry in healthy individuals. Therefore, we expect that other factors than the mask type or the use of a metabolic testing mask may affect the performance of participants.

All the previously cited studies on surgical face masks are concordant and support our findings about the minimal impact of these masks on exercise testing results. Therefore, we conclude that, in most cases, treadmill stress testing for diagnostic and prognostic indications can be interpreted similarly if performed with a surgical face mask or without a mask. In other words, clinicians are encouraged to err on the side of caution and consider that decreased exercise capacity is attributable to physiological changes and not the surgical mask. Further studies are required to investigate the effect of masks in populations with significant disease burden. Moreover, the significant decrease in exercise capacity and increase in perceived exertion may have an impact in other exercise settings.

In our study, even if the difference in perceived exertion was small, dyspnea was evoked as the principal reason to stop the test in a greater proportion of participants when wearing the surgical face mask (64% with mask vs 36% without mask). In Driver’s study, the reported dyspnea was significantly increased during exercise with the cloth face mask compared to without a mask, and minute ventilation was significantly reduced. In Finkenzer’s study, dyspnea was not directly assessed, but discomfort during the effort was more important with both N95 and surgical face masks compared to without a mask. Finally, in Epstein’s study, wearing an N95 face mask compared to no mask during the stress test was associated with a small but statistically significant increase in end-tidal CO2 during most of the test, while wearing a surgical face mask increased end-tidal CO2 only when reaching exhaustion. This end-tidal CO2 increase may exacerbate the sensation of dyspnea. The current data do not allow us to fully elucidate the mechanisms and anticipate the clinical repercussions of the increased dyspnea associated with wearing different types of masks. Even if this dyspnea does not impair significantly exercise capacity during a GXT with a surgical mask, it could potentially have an impact on exercise in different conditions, for example in patients engaging in cardiopulmonary rehabilitation or patients with a severe cardiac or pulmonary disease.

STUDY LIMITATIONS

The number of participants recruited was small, but our sample had sufficient statistical power to detect clinically relevant changes. However, the recruitment methods may have induced a bias in the population because an important proportion of participants were healthcare workers used to wearing surgical face masks for prolonged periods. This choice was made to protect the study personnel by including only participants vaccinated against COVID-19. Additionally, this study population included only healthy volunteers. Therefore, further investigation is needed to validate the applicability of our results to patients with cardiovascular or pulmonary conditions. However, as shown in the studies with pulmonary testing during exercise, the principal effects of the mask are observed in the ventilation and respiratory rate. We could hypothesize that most of the patients with cardiovascular diseases will be limited during stress testing by the cardiac output component of their global fitness before the ventilation and consequently that the impact of the mask on their exercise capacity would be like the one observed in our study. However, exercise physiology can be altered in different ways by cardiovascular or respiratory diseases, and to our knowledge, no randomized study assessed the effect of wearing a mask during a GTX in these populations.

The maximal VO2 values (expressed in METs) reported in this study were estimated rather than measured. This reflects our clinical practice, in which stress testing by cardiologists does not include ventilation or respiratory gas parameters. However, potential errors caused by this approach should be the same during testing with and without a mask and will not impact our conclusions.

CONCLUSION

The 0.4 MET decrease in effort capacity and 0.5-point increase on the Borg scale of perceived exertion found in this study for healthy participants undergoing GXT while wearing a surgical face mask are statistically significant but not clinically significant. However, dyspnea is more frequently reported as the reason to stop the test in participants wearing masks. Based on these findings, GXT should be interpreted in the same manner regardless of whether participants wore a surgical face mask or no mask. However, for people with cardiovascular or pulmonary disease presenting with dyspnea, the differences may be more pronounced and require further investigation.